Cursor's negative 23% gross margins revealed ahead of reported $60B acquisition talks

Cursor is generating substantial revenue but operating at negative 23% gross margins, according to financial details revealed ahead of reported $60 billion acquisition talks with SpaceX.

The margin figure is notably poor for a company at Cursor's revenue scale. The critical question is whether those margins will improve once Cursor's own trained models power the inference layer instead of relying on third-party API costs that currently compress profitability.

The stickiness angle matters here. Cursor users have shown willingness to stay on the platform while switching between inference providers based on what their subscription plan offers. They are subscribers first—willing to work within budget constraints by choosing different models. That behavior suggests Cursor's value proposition sits upstream of any single inference provider, which could be defensible.

The SpaceX and xAI deals hint at a path forward. If Cursor can merge its lessons from Composer (its agent-based editing tool) and Cryc with models trained internally at xAI or xAI-aligned infrastructure, the economics could shift. Higher-margin inference running on owned or affiliated compute would directly improve unit economics. Whether that materializes remains uncertain, but the mechanism is straightforward: margin compression today reflects Cursor's dependence on expensive third-party inference; margin expansion requires controlling the inference layer.

The buyout narrative frames this acquisition not as a distressed sale but as a scaling opportunity. At negative 23% gross margins, Cursor isn't broken—it's capital-intensive. A well-capitalized acquirer with aligned compute infrastructure could theoretically make the model work.